1 /* Definitions for CPP library.
2 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001
3 Free Software Foundation, Inc.
4 Written by Per Bothner, 1994-95.
6 This program is free software; you can redistribute it and/or modify it
7 under the terms of the GNU General Public License as published by the
8 Free Software Foundation; either version 2, or (at your option) any
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
20 In other words, you are welcome to use, share and improve this program.
21 You are forbidden to forbid anyone else to use, share and improve
22 what you give them. Help stamp out software-hoarding! */
26 #include <sys/types.h>
27 #include "hashtable.h"
33 /* For complex reasons, cpp_reader is also typedefed in c-pragma.h. */
34 #ifndef GCC_C_PRAGMA_H
35 typedef struct cpp_reader cpp_reader;
37 typedef struct cpp_buffer cpp_buffer;
38 typedef struct cpp_options cpp_options;
39 typedef struct cpp_token cpp_token;
40 typedef struct cpp_string cpp_string;
41 typedef struct cpp_hashnode cpp_hashnode;
42 typedef struct cpp_macro cpp_macro;
43 typedef struct cpp_lexer_pos cpp_lexer_pos;
44 typedef struct cpp_lookahead cpp_lookahead;
45 typedef struct cpp_callbacks cpp_callbacks;
48 struct file_name_map_list;
51 /* The first two groups, apart from '=', can appear in preprocessor
52 expressions. This allows a lookup table to be implemented in
55 The first group, to CPP_LAST_EQ, can be immediately followed by an
56 '='. The lexer needs operators ending in '=', like ">>=", to be in
57 the same order as their counterparts without the '=', like ">>". */
59 /* Positions in the table. */
60 #define CPP_LAST_EQ CPP_MAX
61 #define CPP_FIRST_DIGRAPH CPP_HASH
62 #define CPP_LAST_PUNCTUATOR CPP_DOT_STAR
67 OP(CPP_GREATER, ">") /* compare */ \
69 OP(CPP_PLUS, "+") /* math */ \
74 OP(CPP_AND, "&") /* bit ops */ \
77 OP(CPP_RSHIFT, ">>") \
78 OP(CPP_LSHIFT, "<<") \
79 OP(CPP_MIN, "<?") /* extension */ \
83 OP(CPP_AND_AND, "&&") /* logical */ \
87 OP(CPP_COMMA, ",") /* grouping */ \
88 OP(CPP_OPEN_PAREN, "(") \
89 OP(CPP_CLOSE_PAREN, ")") \
90 OP(CPP_EQ_EQ, "==") /* compare */ \
91 OP(CPP_NOT_EQ, "!=") \
92 OP(CPP_GREATER_EQ, ">=") \
93 OP(CPP_LESS_EQ, "<=") \
95 OP(CPP_PLUS_EQ, "+=") /* math */ \
96 OP(CPP_MINUS_EQ, "-=") \
97 OP(CPP_MULT_EQ, "*=") \
98 OP(CPP_DIV_EQ, "/=") \
99 OP(CPP_MOD_EQ, "%=") \
100 OP(CPP_AND_EQ, "&=") /* bit ops */ \
101 OP(CPP_OR_EQ, "|=") \
102 OP(CPP_XOR_EQ, "^=") \
103 OP(CPP_RSHIFT_EQ, ">>=") \
104 OP(CPP_LSHIFT_EQ, "<<=") \
105 OP(CPP_MIN_EQ, "<?=") /* extension */ \
106 OP(CPP_MAX_EQ, ">?=") \
107 /* Digraphs together, beginning with CPP_FIRST_DIGRAPH. */ \
108 OP(CPP_HASH, "#") /* digraphs */ \
109 OP(CPP_PASTE, "##") \
110 OP(CPP_OPEN_SQUARE, "[") \
111 OP(CPP_CLOSE_SQUARE, "]") \
112 OP(CPP_OPEN_BRACE, "{") \
113 OP(CPP_CLOSE_BRACE, "}") \
114 /* The remainder of the punctuation. Order is not significant. */ \
115 OP(CPP_SEMICOLON, ";") /* structure */ \
116 OP(CPP_ELLIPSIS, "...") \
117 OP(CPP_PLUS_PLUS, "++") /* increment */ \
118 OP(CPP_MINUS_MINUS, "--") \
119 OP(CPP_DEREF, "->") /* accessors */ \
121 OP(CPP_SCOPE, "::") \
122 OP(CPP_DEREF_STAR, "->*") \
123 OP(CPP_DOT_STAR, ".*") \
124 OP(CPP_ATSIGN, "@") /* used in Objective C */ \
126 TK(CPP_NAME, SPELL_IDENT) /* word */ \
127 TK(CPP_INT, SPELL_STRING) /* 23 */ \
128 TK(CPP_FLOAT, SPELL_STRING) /* 3.14159 */ \
129 TK(CPP_NUMBER, SPELL_STRING) /* 34_be+ta */ \
131 TK(CPP_CHAR, SPELL_STRING) /* 'char' */ \
132 TK(CPP_WCHAR, SPELL_STRING) /* L'char' */ \
133 TK(CPP_OTHER, SPELL_CHAR) /* stray punctuation */ \
135 TK(CPP_STRING, SPELL_STRING) /* "string" */ \
136 TK(CPP_WSTRING, SPELL_STRING) /* L"string" */ \
137 TK(CPP_HEADER_NAME, SPELL_STRING) /* <stdio.h> in #include */ \
139 TK(CPP_COMMENT, SPELL_STRING) /* Only if output comments. */ \
140 TK(CPP_MACRO_ARG, SPELL_NONE) /* Macro argument. */ \
141 OP(CPP_EOF, "EOL") /* End of line or file. */
153 /* C language kind, used when calling cpp_reader_init. */
154 enum c_lang {CLK_GNUC89 = 0, CLK_GNUC99, CLK_STDC89, CLK_STDC94, CLK_STDC99,
155 CLK_GNUCXX, CLK_CXX98, CLK_OBJC, CLK_OBJCXX, CLK_ASM};
157 /* Payload of a NUMBER, FLOAT, STRING, or COMMENT token. */
161 const unsigned char *text;
164 /* Flags for the cpp_token structure. */
165 #define PREV_WHITE (1 << 0) /* If whitespace before this token. */
166 #define DIGRAPH (1 << 1) /* If it was a digraph. */
167 #define STRINGIFY_ARG (1 << 2) /* If macro argument to be stringified. */
168 #define PASTE_LEFT (1 << 3) /* If on LHS of a ## operator. */
169 #define NAMED_OP (1 << 4) /* C++ named operators. */
170 #define NO_EXPAND (1 << 5) /* Do not macro-expand this token. */
171 #define AVOID_LPASTE (1 << 6) /* Check left for accidental pastes. */
173 /* A preprocessing token. This has been carefully packed and should
174 occupy 12 bytes on 32-bit hosts and 16 bytes on 64-bit hosts. */
177 ENUM_BITFIELD(cpp_ttype) type : CHAR_BIT; /* token type */
178 unsigned char flags; /* flags - see above */
182 cpp_hashnode *node; /* An identifier. */
183 struct cpp_string str; /* A string, or number. */
184 unsigned int arg_no; /* Argument no. for a CPP_MACRO_ARG. */
185 unsigned char c; /* Character represented by CPP_OTHER. */
189 /* The position of a token in the current file. */
193 unsigned int output_line;
197 typedef struct cpp_token_with_pos cpp_token_with_pos;
198 struct cpp_token_with_pos
204 /* Token lookahead. */
207 struct cpp_lookahead *next;
208 cpp_token_with_pos *tokens;
210 unsigned int cur, count, cap;
213 /* A standalone character. We may want to make it unsigned for the
214 same reason we use unsigned char - to avoid signedness issues. */
215 typedef int cppchar_t;
217 /* Values for opts.dump_macros.
218 dump_only means inhibit output of the preprocessed text
219 and instead output the definitions of all user-defined
220 macros in a form suitable for use as input to cpp.
221 dump_names means pass #define and the macro name through to output.
222 dump_definitions means pass the whole definition (plus #define) through
224 enum { dump_none = 0, dump_only, dump_names, dump_definitions };
226 /* This structure is nested inside struct cpp_reader, and
227 carries all the options visible to the command line. */
230 /* Name of input and output files. */
231 const char *in_fname;
232 const char *out_fname;
234 /* Characters between tab stops. */
235 unsigned int tabstop;
237 /* Pending options - -D, -U, -A, -I, -ixxx. */
238 struct cpp_pending *pending;
240 /* File name which deps are being written to. This is 0 if deps are
241 being written to stdout. */
242 const char *deps_file;
244 /* Search paths for include files. */
245 struct search_path *quote_include; /* "" */
246 struct search_path *bracket_include; /* <> */
248 /* Map between header names and file names, used only on DOS where
249 file names are limited in length. */
250 struct file_name_map_list *map_list;
252 /* Directory prefix that should replace `/usr/lib/gcc-lib/TARGET/VERSION'
253 in the standard include file directories. */
254 const char *include_prefix;
255 unsigned int include_prefix_len;
257 /* -fleading_underscore sets this to "_". */
258 const char *user_label_prefix;
260 /* The language we're preprocessing. */
263 /* Non-0 means -v, so print the full set of include dirs. */
264 unsigned char verbose;
266 /* Nonzero means use extra default include directories for C++. */
267 unsigned char cplusplus;
269 /* Nonzero means handle cplusplus style comments */
270 unsigned char cplusplus_comments;
272 /* Nonzero means handle #import, for objective C. */
275 /* Nonzero means don't copy comments into the output file. */
276 unsigned char discard_comments;
278 /* Nonzero means process the ISO trigraph sequences. */
279 unsigned char trigraphs;
281 /* Nonzero means process the ISO digraph sequences. */
282 unsigned char digraphs;
284 /* Nonzero means to allow hexadecimal floats and LL suffixes. */
285 unsigned char extended_numbers;
287 /* Nonzero means print the names of included files rather than the
288 preprocessed output. 1 means just the #include "...", 2 means
289 #include <...> as well. */
290 unsigned char print_deps;
292 /* Nonzero if phony targets are created for each header. */
293 unsigned char deps_phony_targets;
295 /* Nonzero if missing .h files in -M output are assumed to be
296 generated files and not errors. */
297 unsigned char print_deps_missing_files;
299 /* If true, fopen (deps_file, "a") else fopen (deps_file, "w"). */
300 unsigned char print_deps_append;
302 /* Nonzero means print names of header files (-H). */
303 unsigned char print_include_names;
305 /* Nonzero means cpp_pedwarn causes a hard error. */
306 unsigned char pedantic_errors;
308 /* Nonzero means don't print warning messages. */
309 unsigned char inhibit_warnings;
311 /* Nonzero means don't suppress warnings from system headers. */
312 unsigned char warn_system_headers;
314 /* Nonzero means don't print error messages. Has no option to
315 select it, but can be set by a user of cpplib (e.g. fix-header). */
316 unsigned char inhibit_errors;
318 /* Nonzero means warn if slash-star appears in a comment. */
319 unsigned char warn_comments;
321 /* Nonzero means warn if there are any trigraphs. */
322 unsigned char warn_trigraphs;
324 /* Nonzero means warn if #import is used. */
325 unsigned char warn_import;
327 /* Nonzero means warn about various incompatibilities with
329 unsigned char warn_traditional;
331 /* Nonzero means turn warnings into errors. */
332 unsigned char warnings_are_errors;
334 /* Nonzero causes output not to be done, but directives such as
335 #define that have side effects are still obeyed. */
336 unsigned char no_output;
338 /* Nonzero means we should look for header.gcc files that remap file
342 /* Nonzero means don't output line number information. */
343 unsigned char no_line_commands;
345 /* Nonzero means -I- has been seen, so don't look for #include "foo"
346 the source-file directory. */
347 unsigned char ignore_srcdir;
349 /* Zero means dollar signs are punctuation. */
350 unsigned char dollars_in_ident;
352 /* Nonzero means warn if undefined identifiers are evaluated in an #if. */
353 unsigned char warn_undef;
355 /* Nonzero for the 1999 C Standard, including corrigenda and amendments. */
358 /* Nonzero means give all the error messages the ANSI standard requires. */
359 unsigned char pedantic;
361 /* Nonzero means we're looking at already preprocessed code, so don't
362 bother trying to do macro expansion and whatnot. */
363 unsigned char preprocessed;
365 /* Nonzero disables all the standard directories for headers. */
366 unsigned char no_standard_includes;
368 /* Nonzero disables the C++-specific standard directories for headers. */
369 unsigned char no_standard_cplusplus_includes;
371 /* Nonzero means dump macros in some fashion - see above. */
372 unsigned char dump_macros;
374 /* Nonzero means pass #include lines through to the output. */
375 unsigned char dump_includes;
377 /* Print column number in error messages. */
378 unsigned char show_column;
380 /* Treat C++ alternate operator names special. */
381 unsigned char operator_names;
383 /* True if --help, --version or --target-help appeared in the
384 options. Stand-alone CPP should then bail out after option
385 parsing; drivers might want to continue printing help. */
386 unsigned char help_only;
388 /* SDCC abuse by Kevin: allow naked '#' characters in expanded macros
389 * (see _cpp_create_definition in cppmacro.c)
391 unsigned char allow_naked_hash;
394 /* This structure is passed to the call back when changing file. */
395 enum cpp_fc_reason {FC_ENTER = 0, FC_LEAVE, FC_RENAME};
399 const char *filename;
403 typedef struct cpp_file_change cpp_file_change;
404 struct cpp_file_change
406 struct cpp_file_loc from; /* Line of #include or #line. */
407 struct cpp_file_loc to; /* Line after #include or #line, or start. */
408 enum cpp_fc_reason reason; /* Reason for change. */
409 unsigned char sysp; /* Nonzero if system header. */
410 unsigned char externc; /* Nonzero if wrapper needed. */
416 void (*file_change) PARAMS ((cpp_reader *, const cpp_file_change *));
417 void (*include) PARAMS ((cpp_reader *, const unsigned char *,
419 void (*define) PARAMS ((cpp_reader *, cpp_hashnode *));
420 void (*undef) PARAMS ((cpp_reader *, cpp_hashnode *));
421 void (*poison) PARAMS ((cpp_reader *));
422 void (*ident) PARAMS ((cpp_reader *, const cpp_string *));
423 void (*def_pragma) PARAMS ((cpp_reader *));
426 #define CPP_FATAL_LIMIT 1000
427 /* True if we have seen a "fatal" error. */
428 #define CPP_FATAL_ERRORS(PFILE) (cpp_errors (PFILE) >= CPP_FATAL_LIMIT)
430 /* Name under which this program was invoked. */
431 extern const char *progname;
433 /* Where does this buffer come from? A faked include, a source file,
434 a builtin macro, a command-line option, or a _Pragma operator. */
435 enum cpp_buffer_type {BUF_FAKE, BUF_FILE, BUF_BUILTIN,
436 BUF_CL_OPTION, BUF_PRAGMA};
438 /* The structure of a node in the hash table. The hash table has
439 entries for all identifiers: either macros defined by #define
440 commands (type NT_MACRO), assertions created with #assert
441 (NT_ASSERTION), or neither of the above (NT_VOID). Builtin macros
442 like __LINE__ are flagged NODE_BUILTIN. Poisioned identifiers are
443 flagged NODE_POISONED. NODE_OPERATOR (C++ only) indicates an
444 identifier that behaves like an operator such as "xor".
445 NODE_DIAGNOSTIC is for speed in lex_token: it indicates a
446 diagnostic may be required for this node. Currently this only
447 applies to __VA_ARGS__ and poisoned identifiers. */
449 /* Hash node flags. */
450 #define NODE_OPERATOR (1 << 0) /* C++ named operator. */
451 #define NODE_POISONED (1 << 1) /* Poisoned identifier. */
452 #define NODE_BUILTIN (1 << 2) /* Builtin macro. */
453 #define NODE_DIAGNOSTIC (1 << 3) /* Possible diagnostic when lexed. */
454 #define NODE_WARN (1 << 4) /* Warn if redefined or undefined. */
456 /* Different flavors of hash node. */
459 NT_VOID = 0, /* No definition yet. */
460 NT_MACRO, /* A macro of some form. */
461 NT_ASSERTION /* Predicate for #assert. */
464 /* Different flavors of builtin macro. */
467 BT_SPECLINE = 0, /* `__LINE__' */
468 BT_DATE, /* `__DATE__' */
469 BT_FILE, /* `__FILE__' */
470 BT_BASE_FILE, /* `__BASE_FILE__' */
471 BT_INCLUDE_LEVEL, /* `__INCLUDE_LEVEL__' */
472 BT_TIME, /* `__TIME__' */
473 BT_STDC /* `__STDC__' */
476 #define CPP_HASHNODE(HNODE) ((cpp_hashnode *) (HNODE))
477 #define HT_NODE(NODE) ((ht_identifier *) (NODE))
478 #define NODE_LEN(NODE) HT_LEN (&(NODE)->ident)
479 #define NODE_NAME(NODE) HT_STR (&(NODE)->ident)
481 /* The common part of an identifier node shared amongst all 3 C front
482 ends. Also used to store CPP identifiers, which are a superset of
483 identifiers in the grammatical sense. */
486 struct ht_identifier ident;
487 unsigned short arg_index; /* Macro argument index. */
488 unsigned char directive_index; /* Index into directive table. */
489 unsigned char rid_code; /* Rid code - for front ends. */
490 ENUM_BITFIELD(node_type) type : 8; /* CPP node type. */
491 unsigned char flags; /* CPP flags. */
495 cpp_macro *macro; /* If a macro. */
496 struct answer *answers; /* Answers to an assertion. */
497 enum cpp_ttype operator; /* Code for a named operator. */
498 enum builtin_type builtin; /* Code for a builtin macro. */
502 /* Call this first to get a handle to pass to other functions. If you
503 want cpplib to manage its own hashtable, pass in a NULL pointer.
504 Or you can pass in an initialised hash table that cpplib will use;
505 this technique is used by the C front ends. */
506 extern cpp_reader *cpp_create_reader PARAMS ((struct ht *,
509 /* Call this to release the handle. Any use of the handle after this
510 function returns is invalid. Returns cpp_errors (pfile). */
511 extern int cpp_destroy PARAMS ((cpp_reader *));
513 /* Call these to get pointers to the options and callback structures
514 for a given reader. These pointers are good until you call
515 cpp_finish on that reader. You can either edit the callbacks
516 through the pointer returned from cpp_get_callbacks, or set them
517 with cpp_set_callbacks. */
518 extern cpp_options *cpp_get_options PARAMS ((cpp_reader *));
519 extern cpp_callbacks *cpp_get_callbacks PARAMS ((cpp_reader *));
520 extern void cpp_set_callbacks PARAMS ((cpp_reader *, cpp_callbacks *));
522 /* Now call cpp_handle_option[s] to handle 1[or more] switches. The
523 return value is the number of arguments used. If
524 cpp_handle_options returns without using all arguments, it couldn't
525 understand the next switch. When there are no switches left, you
526 must call cpp_post_options before calling cpp_start_read. Only
527 after cpp_post_options are the contents of the cpp_options
528 structure reliable. */
529 extern int cpp_handle_options PARAMS ((cpp_reader *, int, char **));
530 extern int cpp_handle_option PARAMS ((cpp_reader *, int, char **));
531 extern void cpp_post_options PARAMS ((cpp_reader *));
534 extern unsigned int cpp_errors PARAMS ((cpp_reader *));
536 extern unsigned int cpp_token_len PARAMS ((const cpp_token *));
537 extern unsigned char *cpp_token_as_text PARAMS ((cpp_reader *,
539 extern unsigned char *cpp_spell_token PARAMS ((cpp_reader *, const cpp_token *,
541 extern void cpp_register_pragma PARAMS ((cpp_reader *,
542 const char *, const char *,
543 void (*) PARAMS ((cpp_reader *))));
544 extern void cpp_register_pragma_space PARAMS ((cpp_reader *, const char *));
546 extern int cpp_start_read PARAMS ((cpp_reader *, const char *));
547 extern void cpp_finish PARAMS ((cpp_reader *));
548 extern int cpp_avoid_paste PARAMS ((cpp_reader *, const cpp_token *,
550 extern enum cpp_ttype cpp_can_paste PARAMS ((cpp_reader *, const cpp_token *,
551 const cpp_token *, int *));
552 extern void cpp_get_token PARAMS ((cpp_reader *, cpp_token *));
553 extern const cpp_lexer_pos *cpp_get_line PARAMS ((cpp_reader *));
554 extern const unsigned char *cpp_macro_definition PARAMS ((cpp_reader *,
555 const cpp_hashnode *));
557 /* Evaluate a CPP_CHAR or CPP_WCHAR token. */
559 cpp_interpret_charconst PARAMS ((cpp_reader *, const cpp_token *,
560 int, int, unsigned int *));
562 extern void cpp_define PARAMS ((cpp_reader *, const char *));
563 extern void cpp_assert PARAMS ((cpp_reader *, const char *));
564 extern void cpp_undef PARAMS ((cpp_reader *, const char *));
565 extern void cpp_unassert PARAMS ((cpp_reader *, const char *));
567 extern cpp_buffer *cpp_push_buffer PARAMS ((cpp_reader *,
568 const unsigned char *, size_t,
569 enum cpp_buffer_type,
571 extern cpp_buffer *cpp_pop_buffer PARAMS ((cpp_reader *));
572 extern int cpp_defined PARAMS ((cpp_reader *, const unsigned char *, int));
574 /* N.B. The error-message-printer prototypes have not been nicely
575 formatted because exgettext needs to see 'msgid' on the same line
576 as the name of the function in order to work properly. Only the
577 string argument gets a name in an effort to keep the lines from
578 getting ridiculously oversized. */
580 extern void cpp_ice PARAMS ((cpp_reader *, const char *msgid, ...))
582 extern void cpp_fatal PARAMS ((cpp_reader *, const char *msgid, ...))
584 extern void cpp_error PARAMS ((cpp_reader *, const char *msgid, ...))
586 extern void cpp_warning PARAMS ((cpp_reader *, const char *msgid, ...))
588 extern void cpp_pedwarn PARAMS ((cpp_reader *, const char *msgid, ...))
590 extern void cpp_notice PARAMS ((cpp_reader *, const char *msgid, ...))
592 extern void cpp_error_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
594 extern void cpp_warning_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
596 extern void cpp_pedwarn_with_line PARAMS ((cpp_reader *, int, int, const char *msgid, ...))
598 extern void cpp_pedwarn_with_file_and_line PARAMS ((cpp_reader *, const char *, int, int, const char *msgid, ...))
600 extern void cpp_error_from_errno PARAMS ((cpp_reader *, const char *));
601 extern void cpp_notice_from_errno PARAMS ((cpp_reader *, const char *));
604 extern int cpp_ideq PARAMS ((const cpp_token *,
606 extern void cpp_output_line PARAMS ((cpp_reader *, FILE *));
607 extern void cpp_output_token PARAMS ((const cpp_token *, FILE *));
608 extern const char *cpp_type2name PARAMS ((enum cpp_ttype));
609 extern unsigned int cpp_parse_escape PARAMS ((cpp_reader *,
610 const unsigned char **,
611 const unsigned char *,
612 unsigned HOST_WIDE_INT, int));
616 /* Lookup an identifier in the hashtable. Puts the identifier in the
617 table if it is not already there. */
618 extern cpp_hashnode *cpp_lookup PARAMS ((cpp_reader *,
619 const unsigned char *,
622 typedef int (*cpp_cb) PARAMS ((cpp_reader *, cpp_hashnode *, void *));
623 extern void cpp_forall_identifiers PARAMS ((cpp_reader *,
627 extern void cpp_scan_buffer_nooutput PARAMS ((cpp_reader *, int));
628 extern void cpp_start_lookahead PARAMS ((cpp_reader *));
629 extern void cpp_stop_lookahead PARAMS ((cpp_reader *, int));
630 extern int cpp_sys_macro_p PARAMS ((cpp_reader *));
633 extern int cpp_included PARAMS ((cpp_reader *, const char *));
634 extern void cpp_make_system_header PARAMS ((cpp_reader *, int, int));
640 #endif /* ! GCC_CPPLIB_H */